The Pathophysiology of Acute Graft-Versus-Host Disease


Dr M. Jaksch, PhD, Division of Clinical Immunology, F79, Karolinska University Hospital, SE-141 86 Stockholm, Sweden. E-mail:


Despite improvements in allogeneic stem cell transplantation, acute graft-versus-host disease (GVHD) remains a significant problem after transplantation, and it is still a major cause of post-transplant mortality. Disease progression is characterized by the differentiation of alloreactive T cells to effector cells leading to tissue damage, recruitment of additional inflammatory cell populations and further cytokine dysregulation. To make the complex process of acute GVHD more explicit, the pathophysiology of acute GVHD is often divided into three different phases. This review summarizes the mechanisms involved in the three phases of acute GVHD.


Over the past half century, hematopoietic stem cell transplantation (SCT) has evolved from an idea to a well-established therapy used to treat tens of thousands of individuals annually. Hematopoietic cell transplantation is the preferred therapy for a substantial proportion of patients with life-threatening diseases of the lymphohematopoietic system.

One persistent problem of SCT has been graft-versus-host disease (GVHD). Most recipients of an allogeneic SCT experience some degree of acute GVHD after transplantation. This severe reaction is defined as a rapidly progressing systemic illness characterized by immunosuppression and tissue injury in various organs such as the liver, skin and intestinal mucosa [1, 2]. Cell destruction in these areas results in rash, mucosal denudation, subsequent diarrhoea and biliary stasis. Despite progress in understanding the mediators involved in acute GVHD, effective treatment has remained elusive; most patients who develop the severe manifestation of the disease succumb to it or to complications of its treatment. The complex pathophysiology involves host tissue damage, which results from the conditioning regimen (chemotherapy and/or irradiation), inflammatory cytokines and effector cells [3].

The first description of acute GVHD came from rodent experiments documenting hematopoietic reconstitution after marrow lethal radiation [4]. Animals that received syngenic stem cell graft recovered from the radiation toxicity (primary disease) and appeared to be normal. Animals that received their stem cells from different strains [with major histocompatibility complex (MHC) loci differences] recovered from their primary disease but developed secondary disease, which is now known as acute GVHD [5]. In 1966, Billingham outlined the requirements for the induction of GVHD. First, the graft must contain immunologically competent cells. Second, the host must appear foreign to the graft and therefore be capable of stimulating the donor cells. Finally, the host immune system must be incapable of generating an immune response [6]. Herein, we will review the pathophysiology and mechanisms involved in acute GVHD after allogeneic SCT.

Three-phase model of acute GVHD

Although the interactions of subsets of T cells and effector cells during acute GVHD via production of cytokines is a rather complex multistep process, it is by now generally accepted that acute GVHD can be summarized in a three-step process with an afferent and efferent phase [7–9]. Injury to host epithelium and endothelium generates injury signals that recruit donor T cells (Phase I). In this inflammatory milieu, the donor T cells recognize alloantigens, resulting in activation and proliferation (Phase II). The first two phases make up the afferent phase of GVHD. Finally, the T cells cause further injury through both specific and non-specific mechanisms during the efferent phase (Phase III). The three-phase model of acute GVHD is summarized in Fig. 1.

Figure 1.

The three-phase model of acute graft-versus-host disease (GVHD). During step 1, the conditioning regimen (irradiation and/or chemotherapy) leads to damage, activation of host tissues and induction of inflammatory cytokines [tumour necrosis factor (TNF)-α and interleukin (IL)-1] secretion. Increased expression of major histocompatibility complex (MHC) antigens and adhesion molecules leads to enhancement of the recognition of host MHC and/or minor histocompatibility antigens by mature donor T cells. During step 2, donor T cells proliferate and secrete IL-2 and interferon (IFN)-γ. These cytokines induce further T-cell expansion, induce cytotoxic T lymphocytes (CTL) and natural killer (NK) cells responses and prime additional mononuclear phagocytes to produce TNF-α and IL-1. Also, nitric oxide (NO) is produced by activated macrophages, and it may contribute to the tissue damage seen during step 3. Lipopolysaccharide (LPS), which leaks through the intestinal mucosa that was damaged during step 1, together with IFN-γ, from step 2, further stimulate macrophages to secrete cytokines and NO. During step 3, CTL and NK cells induce target tissue damage through cell-mediated cytotoxicity.

Phase I: Effects of conditioning

The fist step of acute GVHD occurs during the conditioning phase, a phase that occurs even before donor cells are infused into the host. The conditioning therapy, which includes irradiation and chemotherapy, is rather toxic to tissues and leads to damage and activation of host tissues, including intestinal mucosa, liver and other tissues. The injured tissues will respond with the production of factors – such as cytokines, chemokines and adhesion molecules – that signal to the immune system that injury have occurred [10]. Hence, donor T cells are infused into a host whose tissues have already been damaged by several factors such as the patient's underlying disease and its earlier treatment, infections and preparative conditioning consisting of high-dose therapy given before SCT. The most commonly secreted inflammatory cytokines that are secreted by activated host cells are tumour necrosis factor (TNF)-α and interleukin (IL)-1 [11]. The presence of inflammatory cytokines during this phase increases the expression of adhesion molecules, costimulatory molecules and MHC antigens [12, 13]. This leads to the activation of host dendritic cells (DC) and recognition of host MHC and/or minor histocompatibility antigens (miH) by mature donor T cells [14]. TNF-α also contributes to the intestinal injury by affecting the integrity of the gut mucosa directly [15]. Further injury to the barrier results in the release of endotoxin, i.e. lipopolysaccharide (LPS) a cell-wall component of gram-negative bacteria, which can leak through the damaged mucosa. LPS may subsequently trigger gut-associated lymphocytes and macrophages to additional production of TNF-α and IL-1 [16]. Elevated serum levels of LPS have been shown to correlate directly with the degree of intestinal damage occurring after allogeneic SCT [17]. The release of LPS and activation of lymphocytes and macrophages may result in the amplification of local tissue injury and further promotion of an inflammatory response. Indirect support for this concept can be found in experimental transplantation and in some clinical settings where transplantation in low-bacteria environments strikingly limits the risk of GVHD [16–18]. Also, blockade of LPS has shown to prevent GVHD and preserve graft-versus-leukaemia (GVL) effect [19]. Moreover, experimental approaches to prevent GVHD include reducing the damage to the gastrointestinal tract by protection of the mucosal barrier. Two unique factors that can shield the gastrointestinal tract from the toxic effects of conditioning therapy, keratinocyte growth factor (KGF) and IL-11 have been shown in experimental models to decrease gastrointestinal toxicity and reduce acute GVHD [20–23]. Consistent with the hypothesis that protection of the intestinal mucosa could block the inflammatory cascade of GVHD, KGF prevented the translocation of LPS into circulation and resulted in a reduction of systemic TNF-α. Also, KGF probably provides its protective effects on acute GVHD by suppression of pro-inflammatory cytokines such as TNF-α and interferon (IFN)-γ[24]. Unfortunately, IL-11 has proved very toxic, leading to closure of a randomized trial [25]. Also, attempts to block the effects of the GVHD-inducing cytokines TNF-α and IL-1 with antibodies have been attractive, but these studies have met limited clinical success [26–28].

Phase II: Donor T-Cell activation

Presentation of recipient antigens to donor T cells, activation of donor T cells and subsequent proliferation and differentiation of these activated T cells are crucial in the second phase of acute GVHD. After infusion of the graft, donor T cells recognize foreign host antigens presented by activated antigen-presenting cells (APC). The greater the disparity between donor and recipient MHC, the greater the T-cell response. Even single-antigen differences between donor and recipient result in significantly more GVHD than in human leucocyte antigen (HLA)-identical pairs. MHC class I differences stimulate CD8+ T cells, and MHC class II differences stimulate CD4+ T cells [29]. In identical pairs, the donor T cells recognize minor antigen differences.

During the process of donor T-cell activation, APC play a very important role by digesting large proteins into smaller peptides and present them on the surface in association with MHC molecules. In particular, DC are uniquely specialized for the uptake and presentation of antigen to naïve T cells. Immature DC are distributed in tissues, particularly barrier organs, such as skin and bowel, and they are specialized for the uptake of antigens via endocytosis. Before, during and immediately after an allogeneic SCT, hosts are exposed to inflammatory cytokines such as TNF-α and IL-1, pathogen-derived products such as LPS and necrotic cells that are damaged by recipient conditioning. All these can initiate DC maturation, reviewed in [30]. The distribution of APC may explain the unusual target organ distribution of acute GVHD. For example, a study has shown that selective removal of APC from a specific organ may reduce GVHD in that target organ but not in other organs [31]. Zhang and colleagues suggested that the DC and macrophages in the secondary lymphoid tissues are critical for the activation and proliferation of donor CD8+ T cells, whereas APC in the target organ are required for the recruitment of previously activated donor T cells to the tissue. These results suggest that the host APC may not only play a role in the activation phase but also in the recruitment of allogeneic cytotoxic T cells during the effector phase of acute GVHD. Another study by Murai and colleagues [32] demonstrated the importance of the Peyer's patches of the small intestine as key sites of antigen presentation to CD8+ donor T cells. This study showed that if Peyer's patches were absent or if the donor T-cell migration to Peyer's patches was blocked, lethal GVHD did not occur.

After the preparative regimen, transplant recipients are chimeric for APC. They have residual recipient-derived APC that survived the conditioning regimen, and they have donor APC derived from the stem cell graft. Although allogeneic antigens can be presented directly by host-derived and indirectly by donor-derived APC, host-derived APC appear to be critical in inducing GVHD across both miH and MHC mismatches. Murine studies have demonstrated that host APC alone are both necessary and sufficient to stimulate donor T cells [33–35]. Therefore, recipient APC depletion may be an effective way of decreasing GVHD induced by CD8+ or CD4+ T cells in MHC-disparate and -identical transplants. In a recent study, Shlomchik et al. extended their earlier findings concerning APC and GVHD and found that also donor-derived APC are important and may intensify the GVHD after SCT. Interestingly, they also demonstrated that donor APC were not required for the desired GVL effect [36]. In addition, the importance of host APC has recently been confirmed in another interesting study [37]. Merad et al. showed that persistent host Langerhans cells, the major APC in the skin, are responsible for cutaneous GVHD. Moreover, the authors reported that ultraviolet irradiation could deplete host Langerhans cells in mice, a depletion which was shown to be protective against GVHD in the skin. Application of these findings to the clinic could have major implications for the prevention of both acute and chronic GVHD and may increase the safety and applicability of SCT.

Donor T-cell activation

Engagement of the T-cell receptor by peptide presented on a MHC molecule of an APC provides the initiating signal for T-cell stimulation. However, a second costimulatory signal is also required for full T-cell stimulation. The outcome of the first signal is regulated by the second signal. Three outcomes may occur: complete activation, partial activation or anergy, i.e. a long-lasting state of antigen-specific unresponsiveness. A growing number of T-cell costimulatory pathways have been identified [38–40]. To date, the most important pathways appear to be mediated by interactions between CD28 with B7 and CD40 with its ligand CD154. In the most completely characterized interaction, B7 binds to T-cell surface receptors CD28 and cytotoxic T-lymphocyte antigen 4 (CTLA-4). CD28 provides a positive signal that lowers the threshold for T-cell activation and promotes T-cell differentiation and survival, whereas CTLA-4 delivers an inhibitory signal [41].

A novel approach to reducing acute GVHD involves the blockade of the T-cell costimulatory signals [42]. In order to block these interactions, either ex vivo manipulation of the donor T cells or systemic administration of blocking agents (e.g. the fusion protein CTLA4-Ig or antibodies) in vivo has been used. In one animal study on GVHD, blocking of the B7–CD28 interaction inhibited both acute and chronic GVHD [43]. Other studies on CD28–/– T cells, CTLA-4-Ig and anti-B7 antibodies have been performed; however, because of variability in strain pairing and transplant conditions, it is difficult to interpret the results. Although blockade of costimulatory pathways seem to be a promising way to prevent acute GVHD, this strategy needs further investigation.


IL-2 and IFN-γ.  The T-cell activation and proliferation is followed by cytokine and chemokine secretion [44]. The activation involves multiple pathways, which activate transcription of genes for cytokines, such as IL-2, IFN-γ and their receptors [45]. IL-2 has long been considered to be the primary cytokine involved in acute GVHD both because of its central role as a T-cell growth factor, and because cyclosporine, a powerful prophylactic agent against acute GVHD, is known to inhibit IL-2 secretion [46, 47]. Under the influence of IL-2 and other immune mediators, alloreactive T cells expand clonally and differentiate into cytotoxic T lymphocytes (CTL). The importance of IL-2 in the pathophysiology of acute GVHD has been shown by administration of monoclonal antibodies against the IL-2 receptor. In animal and clinical studies, administration of monoclonal antibodies early after SCT has in some studies shown to prevent acute GVHD [48, 49]. IL-2 has been shown to control and amplify the immune response against alloantigens. When low doses of IL-2 have been administered after allogeneic SCT, the severity and mortality of GVHD have been enhanced [50, 51]. However, IL-2 appears to play a rather complex role in GVHD. On one hand, neutralization of IL-2 has resulted in the amelioration of GVHD; on the other hand, administration of high doses of IL-2 has shown to inhibit GVHD [52].

IFN-γ is another crucial cytokine involved in the second phase of acute GVHD. It has been shown that IFN-γ levels are significantly higher in mice with acute GVHD than in those without the disease [53]. Together with IL-2, IFN-γ induces further T-cell expansion, induces CTL and natural killer (NK)-cell responses and prime additional mononuclear phagocytes to produce IL-1 and TNF-α. Several experimental studies have shown multiple effects of IFN-γ in the pathophysiology of acute GVHD. As well as other inflammatory cytokines, IFN-γ has shown to induce increased expression of adhesion molecules, chemokines and HLA molecules, which enhances the recruitment of cells and antigen presentation. Another important effect in the context of acute GVHD is the direct tissue damage caused by IFN-γ[54, 55]. It is generally known that the immune system is itself a GVHD target and immunosuppression is a common feature seen during acute GVHD. In several experimental studies, IFN-γ appears to mediate this form of immunosuppression through the induction of nitric oxide (NO) and Fas expression [56, 57]. Also, by enhancement of Fas-mediated apoptosis, IFN-γ plays an important role in regulating cell death of activated donor T cells [58]. Furthermore, IFN-γ also reduces the amount of LPS needed to stimulate macrophages to produce inflammatory cytokines and NO. Interestingly, in this context, IFN-γ has two opposing functions. On one hand, it can mature DC, prime macrophages to produce inflammatory cytokines, and induce NO secretion, all of which intensify the acute GVHD reaction. On the other hand, IFN-γ can decrease GVHD by inducing the expression of Fas receptors on donor T cells, causing activation-induced cell death and diminishing the donor T-cell response to host antigens [59]. It is well known that IFN-γ promotes local inflammation; however, at the systemic level, it initiates an anti-inflammatory response [60]. Hence, IFN-γ can show both a suppressive and stimulating effect under different circumstances.

IL-18 and granulocyte colony-stimulating factor. IL-18 is a recently discovered cytokine that also influences the pathophysiology of acute GVHD. It is produced by a variety of cells, and the major targets of IL-18 include macrophages, NK cells, T cells and B-cells. IL-18 has the capacity of influencing both Th1- and Th2-mediated responses. IL-18 was found to be elevated in acute GVHD, but surprisingly, blockade of IL-18 accelerated acute GVHD mortality in animal models [61]. In the study by Reddy and colleagues, it was also discovered that the administration of IL-18 early after SCT increased the serum levels of IFN-γ, which led to increased expression of Fas receptors on donor CD4+ T cell. This resulted in a reduction of CD4+-cell-mediated acute GVHD by induction of Fas-mediated apoptosis of donor T cells. Administration of IL-18 to the stem cell donor before SCT appeared also to be protective against acute GVHD; however, most likely by an opposite effect, the enhancement of Th2 cytokine production. Furthermore, IL-18 seems to have a rather complex involvement in the biology of acute GVHD. Min and colleagues recently showed that IL-18 seems to have a paradoxical effect on CD4+- and CD8+-cell-mediated GVHD. While administration of IL-18 significantly increased the survival in CD4+-mediated GVHD, the survival was reduced in the CD8+-mediated GVHD model [62]. This study suggests that not only the timing of IL-18 administration and the inflammatory milieu may be critical to the eventual outcome of acute GVHD but also which T-cell subsets that are involved in mediating GVHD.

Peripheral blood stem cells are an alternative source of stem cells for allogeneic SCT. Interestingly, it has been demonstrated that granulocyte colony-stimulating factor (G-CSF)-mobilized donor peripheral blood stem cells (PBSC) reduces the early mortality in acute GVHD after allogeneic SCT in mice [63]. It has been suggested that the reason for this is that pretreatment of donor cells with G-CSF may polarize donor T cells towards Th2 [64, 65]. However, when G-CSF is administered to the patient shortly after SCT in order to shorten the neutrophenic phase, it has been shown to increase the risk of GVHD and death [66].


A characteristic feature of all inflammatory reactions is the excessive recruitment of leucocytes to the site of inflammation. The process of leucocyte recruitment to the target tissue is well-orchestrated and involves several protein families, including pro-inflammatory cytokines, adhesion molecules, matrix metalloproteinases and the large cytokine subfamily of chemotactic cytokines, the chemokines [67, 68]. Inflammatory chemokines are expressed in inflamed tissues by infiltrated cells, monocytes or macrophages or by resident cells, epithelial, endothelial or fibroblastic cells on stimulation by pro-inflammatory cytokines (e.g. IL-1, TNF-α or IFN-γ) or stimuli (e.g. LPS). This group of chemokines is specialized for the recruitment of effector cells, including monocytes, granulocytes and effector T cells. Studies using murine models of acute GVHD have demonstrated the critical role of several chemokines and their receptors (particularly MIP-1α, MIP-2, Mig, MCP-1, MCP-3 and CCR5) by directing T-cell infiltration into target tissues during acute GVHD [69–71]. It has earlier been shown that CCR5-expressing T lymphocytes are recruited to the liver during acute GVHD in mice models [69] and that MIP-1α, a ligand for CCR1 and CCR5, also seems to be involved in liver GVHD [71]. Recently, Duffner and colleagues [72] showed that the migration of donor CD8+ T cells to GVHD target organs such as the intestines depends on the expression of CXCR3 and that the presence of this receptor significantly contributed to GVHD damage and overall mortality in mice. The role of various chemokines and their receptors in regulating donor T-cell migration to GVHD target tissues in clinical SCT remains unexplored. However, our group has recently shown that increased gene expression of CCR5, CXCR3, CCR1 and CCR2 is seen in connection with acute GVHD after allogeneic SCT [73]. Therefore, chemokines and chemokine receptors may not only act as potential targets for modulation of acute GVHD but also as diagnostic markers for early detection of the disease.

NK cells

Although tissue damage in the effector phase of acute GVHD can result from the cytolytic function of CTL, other effector cells such as NK cells seem to be involved in the process. NK cells are negatively regulated by MHC class I-specific inhibitory receptors; thus, HLA-mismatched transplants may trigger donor NK-cell-mediated alloreactivity [74]. NK cells, which reconstitute very rapidly after SCT, can be major producers of IFN-γ, TNF-α and NO upon stimulation, and thus, can contribute to the tissue damage seen in GVHD [75, 76]. However, it has also been suggested that NK cells may suppress GVH reactions and contribute to GVL effects [77, 78]. In murine models of SCT, it has recently been shown that activated donor NK cells prevent GVHD through general elimination of host APC and/or the secretion of the immunosuppressive cytokine TGF-β[78, 79]. In a study by Asai and colleagues, anti-TGF-β completely abrogated the protective effects of activated NK cells, which indicated the important role for TGF-β in the prevention of GVHD by NK cells. However, it is not clear whether NK cells are producing TGF-β or are inducing other cells to make it. The suppressive effect of NK cells on GVHD has also been confirmed in humans. In recent studies, with patients receiving haplo-identical transplants, HLA class I differences, driving donor NK-mediated alloreaction in the GVH direction, mediated potent GVL effects and produced higher engraftment rates without causing acute GVHD [79–81].

Regulatory cells

Recently, a particular subpopulation of CD4+ T cells, which constitutively express the IL-2 receptor α-chain (CD25) and which constitutes 5–10% of the whole CD4+ T-cell pool in mice and humans, has been identified for its crucial role in the control of autoimmune processes [82]. Accumulating evidence has also indicated that regulatory T lymphocytes play an important role in the down-regulation of immune responses to self or allogeneic antigens [83–85]. The mechanism of action of these regulatory T cells is poorly understood and largely controversial. In vitro studies have shown that these cells inhibit the activation of both CD4+ and CD8+ conventional CD25 T cells by acting either directly on the target T cells or on APC [86–89]. In a murine SCT model, Taylor et al. [90] demonstrated that the depletion of CD25+ T cells accelerated the development of GVHD in recipients, an effect that was also seen after in vivo administration of a CD25 monoclonal antibody. Additionally, the infusion of ex vivo-activated and expanded CD4+ CD25+ T cells ameliorated the development of GVHD [90]. Other murine studies have confirmed that infusing donor-derived CD4+ CD25+ T cells may suppress the development of GVHD after allogeneic SCT [91, 92], especially cells that express CD62L [93]. Hoffmann et al. showed that the balance between donor-type CD4+ CD25+ Treg and conventional CD4+ CD25 T cells could determine the outcome of acute GVHD after SCT. Furthermore, they demonstrated not only that IL-10 production by the transplanted Treg cells was necessary for full protection but also that the CD4+ CD25+ Treg cells had to be of donor origin to convey the protection from lethal GVHD [91]. Also, regulatory T cells with a CD3+CD4CD8 (double-negative) phenotype are known to play an important role in preventing the development of GVHD [94]. Interestingly, donor Treg cells seem not to cause generalized immune paralysis, because the beneficial GVL effect of donor T cells has shown to be maintained after SCT in mice [95, 96]. Noteworthy, two recently published papers found that increased frequencies of CD4+CD25+ T cells were present in the donor graft of recipients who experienced acute GVHD [97] and in the peripheral blood of SCT recipients with chronic GVHD [98]. Currently, there is no surface molecule that is truly specific for Treg cells. Hence, these results highlight our need to identify unique markers that may differentiate activated and regulatory CD25-expression T cells. However, it is tempting to speculate that modulating alloimmune responses after SCT with adoptively-transferred donor Treg cells together with specific elimination of alloreactive, activated T cells, as recently described by Martins et al. [99], may be a promising strategy for the prevention or therapy of acute GVHD in humans in the future. With better foresight, and more studies in both mice and men, we might finally advance the therapy for acute GVHD in the future.

Phase III: Cellular and inflammatory effector phase

The third phase of acute GVHD is a complex cascade of multiple effectors. Once donor T cells are activated and proliferate, they are directly or indirectly responsible for the tissue damage seen in GVHD. Three cytolytic pathways are important in the effector function of T cells and other cytolytic cells: the perforin/granzyme B, Fas/Fas ligand (FasL) and direct cytokine-mediated injury. The recent use of knockout mice has demonstrated a central role for each of these pathways in the effector stage of GVHD [1, 100–105].

Cell-mediated cytotoxicity

Although the receptors involved in the recognition of target cells differs between CTL and NK cells, the mechanism by which they kill are essentially the same. They can mediate their cytotoxicity through two different contact-dependent pathways: Perforin-granzyme B-mediated cytolysis and Fas-FasL-mediated apoptosis [103, 106]. However, even though NK cells express FasL and most likely use these molecules to kill certain target cells, NK cells appear to mediate their cytotoxicity primarily through perforin/granzyme-dependent processes [107].

The mechanism of Fas-Fasl and perforin/granzym cytotoxicity. The Fas receptor is a TNF-receptor family member. It is expressed in many tissues, which includes the classic target organs of GVHD, and its level of expression can be increased by pro-inflammatory cytokines during inflammation [108]. The ligand of the Fas receptor (FasL) also belongs to the TNF family and is expressed predominantly on activated T cells, macrophages and neutrophils. Interaction of FasL with the Fas receptor (on the target cell membrane) results in the initiation of the Fas-mediated apoptosis [109].

The pore-forming molecule perforin is another crucial effector molecule of cytolysis by CTL and NK cells. Perforin is expressed mainly by CTL and NK cells and is stored in cytotoxic granules together with granzymes and other proteins, reviewed in [110]. In the presence of calcium, perforin polymerizes and forms channels in the target cell membrane, which allows the granzymes to pass; however, recent in vitro experiments suggest that granzymes may sometimes enter target cells without passing through a perforin channel, but the relative importance of this pathway in vivo is currently unknown [111]. After entering the target cell, granzymes activate caspase cascades, leading to apoptotic cell death [112].

Fas/FasL and perforin/granzyme in acute GVHD. Several studies have shown that the expression of both Fas and FasL is increased on CD8+ and CD4+ donor T cells during acute GVHD in patients and in mice [113–117] and that serum levels of soluble FasL and Fas correlate with the severity of GVHD [118, 119]. In experimental mouse models, the role of the Fas-FasL pathways in the development of GVHD have been analysed by using mice that are deficient for FasL (gld mice) as donors. The gld mice models show that there is a close relationship between the Fas/FasL system and acute GVHD, especially hepatic and cutaneous acute GVHD [1, 103, 104]. Fas-deficient recipients have also been shown to be protected from hepatic GVHD but not from GVHD in other target organs [120]. In addition, the administration of anti-FasL antibodies significantly delays but does not completely reduce the mortality of GVHD [102]. Simultaneous administration of antibodies against FasL and TNF-α completely protected the mice from GVHD. In this study, Hattori et al. verified earlier studies that hepatic GVHD is predominantly mediated by FasL, intestinal GVHD is mainly mediated by TNF-α; and cutaneous GVHD, weight loss, and mortality are mediated by both FasL and TNF-α.

Several groups have created mice deficient for the perforin, granzyme A or B genes [106, 121–124]. These mice grow and develop normally, and their T cells still have the ability to undergo activation. Studies using perforin-deficient donor T cells in various murine SCT models with disparity for MHC class I [101], class II [125] and miH [1] have demonstrated improved survival, an indication that GVHD activity can be mediated through the perforin pathway. Interestingly, the use of these donor T cells did not result in diminished GVHD target organ abnormalities of liver, skin and intestines [1]. The study by Graubert and colleagues also provided evidence that the perforin/granzym pathway is required for class I-restricted GVHD and that FasL is an important mediator of class II-restricted GVHD. However, CD4+ and CD8+ T cells are not restricted to the use of only one cytolytic pathway [105, 125]. Schmaltz et al. also demonstrated that the perforin pathway was important for GVL activity. The importance for the perforin pathway in the GVL effect has also been shown by others [104]. Experimental murine models also suggest that granzyme B-deficient CD8+ T cells have significantly diminished GVHD-induction capability compared with wild-type controls [126].

Inflammatory effectors

Donor T-cell-derived TNF-α.  Although most of the cytolytic activity of CTL can be accounted for by the classic pathways of perforin/granzyme and Fas/FasL, CTL deficient for both pathways exhibit residual cytolytic activity. Braun and colleagues [100] demonstrated that mice that received T cells from donors that were homozygous for non-functional perforin and FasL genes did not develop lethal GVHD; however, the CTL derived from the donors could still display some lytic activity. It has been suggested that TNF, which can be expressed and secreted by activated CTL, could contribute to CTL-mediated cytotoxicity [127]. A role for TNF in the pathogenesis of GVHD has been well documented, but most studies have indicated that GVHD-associated TNF is derived mainly from monocytes and macrophages of donor or host origin [128]. However, the remaining lytic activity by T cells deficient for FasL and perforin has been ascribed to TNF in its membrane-anchored or -secreted form. One group has recently found evidence for a significant contribution of donor T-cell-derived TNF to morbidity and mortality from GVHD as well as to GVL activity [129].

Macrophage-secreted TNF-α and IL-1. In addition to contact-dependent cytotoxicity secretion of inflammatory cytokines, activated macrophages play a key role in causing tissue damage during the third phase of acute GVHD. Mononuclear phagocytes, primed by IFN-γ, are stimulated by LPS to secrete the inflammatory cytokines TNF-α and IL-1. The central role of cytokines as mediators of acute GVHD has recently been demonstrated in a murine model. In this study, severe acute GVHD occurred even in the absence of host alloantigen expression on host target tissues [34].

TNF-α is an inflammatory cytokine that causes a wide variety of biological effects. It activates DC and enhances alloantigen presentation. By inducing inflammatory chemokines, it recruits effector T cells, neutrophils and monocytes into target organs. TNF-α causes direct tissue damage by inducing necrosis of target cells, and it can also induce tissue destruction through apoptosis [57]. It has been shown that serum levels of TNF-α are increased in patients undergoing GVHD after allogeneic SCT [130, 131] and that administration of anti-TNF-α antibodies markedly reduce the weight loss and mortality in a mouse model of acute GVHD [104, 132]. Because TNF-α is thought to be involved in both induction and effector phases of GVHD [2], administration of anti-TNF-α antibodies might diminish not only direct cytotoxic activity of TNF-α but also T-cell activation responsible for acute GVHD. Some beneficial effects of an anti-TNF-α monoclonal antibody (MoAb) for the treatment of refractory acute GVHD have been obtained in the phase I–II clinical trials, but unfortunately GVHD recurred when therapy was discontinued [26].

The second important cytokine that appears to play an important role in the effector phase of acute GVHD is IL-1. The importance of this cytokine has been verified in mice studies, where mice receiving IL-1 after allogeneic SCT displayed an increased mortality that appeared to be an accelerated form of GVHD [133]. Increased gene expression of IL-1 in mononuclear cells has also been shown during clinical acute GVHD [134]. In fact, the use of an IL-1 receptor antagonist (IL-1ra) has shown to reduce acute GVHD in mice models [135, 136]. However, in a recent randomized trial, IL-1ra treatment to prevent acute GVHD was not successful [25].

Nitric oxide. NO is a short-lived biological mediator that plays an important role in host defense and the antimicrobial and tumouricidal function of macrophages. During the development of acute GVHD, increased production of IFN-γ, combined with entry and accumulation of LPS, results in macrophage activation and release of inflammatory products including TNF-α, NO and IL-1. In addition, exposure to increasing amounts of IFN-γ results in a significant reduction in the amount of LPS needed to trigger macrophage synthesis of inflammatory products [137, 138]. As a result of IFN-γ production during the development of acute GVHD, macrophages become primed; therefore, normally insignificant quantities of LPS trigger production of NO and TNF-α[16, 139]. In human and experimental animal transplant recipient, the symptoms of GVHD are preceded by an increase in the serum levels of NO [140, 141]. NO is involved in the effector arm of acute GVHD by inducing immunosuppression and by inhibiting repair mechanisms of target tissue through inactivation of non-heme iron-containing enzymes. This results in the inhibition of proliferation of epithelial stem cells in the gut and skin [142–144] and direct tissue damage [145].

The graft-versus-leukaemia (GVL) effect

GVHD is not only affecting allogeneic SCT in a negative way. There is also a positive side to this reaction, the GVL effect. It is well established that alloreactive donor T lymphocytes react against both the patient's normal hematopoietic cells leading to GVHD and to leukemic cells, known as GVL [146]. These findings have been further confirmed by the fact that lymphocytes of a transplant donor can prevent tumour growth in the recipient by providing donor lymphocyte infusion to induce remission in transplanted patients who have relapsed with chronic myelogenous leukaemia[147–149]. Also, the importance of T cells in achieving long-term engraftment and in effecting graft-versus-tumour reaction has been shown in patients given T-cell-depleted grafts [150, 151]. While these patients had less GVHD, they also had profoundly higher relapse rates.

Because GVHD and GVL are intimately associated, it can be assumed that similar mechanisms are involved in mediating these two reactions. The studies mentioned above all suggest that donor-derived T cells play a central role in the GVL effect. However, donor T cells may not be the only important effector cells involved in the GVL reactivity. NK cells are among the first immune cells to recover after SCT [152], and they mediate cytotoxic effects without prior sensitization. In vivo studies in murine models have shown that transplantation of grafts depleted of T cells but retaining NK cells correlated with reduced relapse rates and minimal incidence of GVHD [153]. However, it remains to be seen whether the graft-versus-malignancy response can truly be separated from more generalized alloreactivity (i.e. GVHD) observed clinically. However, indirect evidence for a GVL effect separated from GVHD in patients with acute leukaemia has been reported by the European Group of Blood and Marrow Transplantation (EBMT) Acute Leukemia Working Party [154] as well as the International Bone Marrow Transplant Registry (IBMTR) [155]. So far, patients with acute GVHD grade I show the highest leukaemia-free survival after SCT [156].

Although donor T cells are likely to be important effector cells for GVL, the target antigens on the tumour cells remain poorly defined. Identifying these targets is of critical importance, because understanding the mechanism of tumour-cell recognition may help to explain why some leukaemias are susceptible to GVL induction and others are not; this knowledge will help in designing better strategies to manipulate the GVL effect for clinical benefit.


GVHD is a potentially devastating consequence of hematopoietic transplantation that is immunologically mediated. It is clear that GVHD is a complex process that is unlikely to be controllable with a single agent. Therefore, a useful strategy will be to attempt to control GVHD by recognizing the underlying pathophysiology and interfering with the three separate phases discussed in this review. For instance, if strategies to maintaining gut integrity, preventing cytokine up-regulation by endotoxin and interfering with T-cell activation are chosen wisely, it may be possible to control the inflammatory aspect of GVHD without loss of the desired GVL effect.

As our understanding of the various factors involved in GVHD advances and as more of these modulators become available for clinical use, real progress will be made in combating this disease. This would in turn make allogeneic SCT more available to patients not considered for SCT today.